Very brief opportunistic interventions for smoking cessation are effective, cost-saving for health systems, and universally recommended in guidelines. However, evidence suggests that clinicians are reluctant to intervene, citing interactional difficulties. Only one UK study has specifically examined smoking discussions, within naturally occurring primary care consultations. However smoking cessation treatment was not available at the time. We examined existing datasets amounting to 519 video-recordings of GP consultations in England for instances of talk about smoking. We used conversation analytic methods to assess patients' responses to doctors asking about smoking, giving advice on smoking, and offering cessation treatment. In 31 recordings it was apparent that the patient smoked, and, in 25/31 consultations, doctors initiated the topic of smoking. They did so by asking about smoking status, commonly during the history-taking phase of the consultation. In many instances, these questions led to active resistance from patients against being placed in a discreditable category, for example by minimising their smoking. This was more pronounced when GPs pursued efforts to quantify the amount smoked. Thereafter, where doctors returned to the topic of smoking, they did so typically by linking smoking to the patient's medical condition, which likewise led to resistance. Guidance recommends that GPs advise on how best to quit smoking where patients are interested in doing so, but this was only evident in a minority of consultations. Where GPs offered support for cessation, they did so using interactional practices that minimised the need for the patient to respond and thereby accept. Interactional difficulties were found to be common in consultations between GPs and people who smoke when GPs actions aligned with some VBA guidelines. Future research should examine when and how advice on how best to quit, and offers of support, should be delivered within primary care consultations.